[AArch64] Add support for Qualcomm Kryo CPU.

[mcrosier]

Summary

Adds core tuning support for new Qualcomm Kryo core.

Chad

[mcrosier]

Fold everything from D17119 and D17122 into a single file, per Renato's request.

I still need for format the MachineModel, per Tim's request.

[mcrosier]

clang-format.

[rengolin]

I obviously haven't checked the scheduler model, but the rest is pretty standard. Thanks!

[mcrosier]

Attempt to format the machine model .td files.

[mcrosier]

FWIW, I verified schedcover.py look good for the scheduling model. Thanks, Tim.

[mcrosier]

Committed r260686.

lib/Target/AArch64/AArch64.td

 
 //===----------------------------------------------------------------------===//
 // AArch64 Processors supported.
 //
 include "AArch64SchedA53.td"
 include "AArch64SchedA57.td"
 include "AArch64SchedCyclone.td"
 include "AArch64SchedM1.td"
+include "AArch64SchedKryo.td"
 
 def ProcA35     : SubtargetFeature<"a35", "ARMProcFamily", "CortexA35",
                                    "Cortex-A35 ARM processors",
                                    [FeatureFPARMv8,
                                    FeatureNEON,
                                    FeatureCrypto,
                                    FeatureCRC,
                                    FeaturePerfMon]>;

 def ProcExynosM1 : SubtargetFeature<"exynosm1", "ARMProcFamily", "ExynosM1",
                                     "Samsung Exynos-M1 processors",
                                     [FeatureFPARMv8,
                                     FeatureNEON,
                                     FeatureCrypto,
                                     FeatureCRC,
                                     FeaturePerfMon]>;
 
+def ProcKryo    : SubtargetFeature<"kryo", "ARMProcFamily", "Kryo",
+                                   "Qualcomm Kryo processors",
+                                   [FeatureFPARMv8,
+                                   FeatureNEON,
+                                   FeatureCrypto,
+                                   FeatureCRC,
+                                   FeaturePerfMon]>;
+
 def : ProcessorModel<"generic", NoSchedModel, [FeatureFPARMv8,
                                               FeatureNEON,
                                               FeatureCRC,
                                               FeaturePerfMon]>;
 
 // FIXME: Cortex-A35 is currently modelled as a Cortex-A53
 def : ProcessorModel<"cortex-a35", CortexA53Model, [ProcA35]>;
 def : ProcessorModel<"cortex-a53", CortexA53Model, [ProcA53]>;
 def : ProcessorModel<"cortex-a57", CortexA57Model, [ProcA57]>;
 // FIXME: Cortex-A72 is currently modelled as an Cortex-A57.
 def : ProcessorModel<"cortex-a72", CortexA57Model, [ProcA57]>;
 def : ProcessorModel<"cyclone", CycloneModel, [ProcCyclone]>;
 def : ProcessorModel<"exynos-m1", ExynosM1Model, [ProcExynosM1]>;
+def : ProcessorModel<"kryo", KryoModel, [ProcKryo]>;
 
 //===----------------------------------------------------------------------===//
 // Assembly parser
 //===----------------------------------------------------------------------===//
 
 def GenericAsmParserVariant : AsmParserVariant {
   int Variant = 0;
   string Name = "generic";

lib/Target/AArch64/AArch64ISelLowering.cpp

for (MVT Ty : {MVT::v2f32, MVT::v4f32, MVT::v2f64}) {
       setOperationAction(ISD::FCEIL, Ty, Legal);
       setOperationAction(ISD::FRINT, Ty, Legal);
       setOperationAction(ISD::FTRUNC, Ty, Legal);
       setOperationAction(ISD::FROUND, Ty, Legal);
     }
   }
 
   // Prefer likely predicted branches to selects on out-of-order cores.
-  if (Subtarget->isCortexA57())
+  if (Subtarget->isCortexA57() || Subtarget->isKryo())
     PredictableSelectIsExpensive = true;
 }
 
 void AArch64TargetLowering::addTypeForNEON(EVT VT, EVT PromotedBitwiseVT) {
   if (VT == MVT::v2f32 || VT == MVT::v4f16) {
     setOperationAction(ISD::LOAD, VT.getSimpleVT(), Promote);
     AddPromotedToType(ISD::LOAD, VT.getSimpleVT(), MVT::v2i32);
 

lib/Target/AArch64/AArch64InstrInfo.cpp

static bool canBeExpandedToORR(const MachineInstr *MI, unsigned BitSize) {
   uint64_t UImm = Imm << (64 - BitSize) >> (64 - BitSize);
   uint64_t Encoding;
   return AArch64_AM::processLogicalImmediate(UImm, BitSize, Encoding);
 }
 
 // FIXME: this implementation should be micro-architecture dependent, so a
 // micro-architecture target hook should be introduced here in future.
 bool AArch64InstrInfo::isAsCheapAsAMove(const MachineInstr *MI) const {
-  if (!Subtarget.isCortexA57() && !Subtarget.isCortexA53())
+  if (!Subtarget.isCortexA57() && !Subtarget.isCortexA53() &&
+      !Subtarget.isKryo())
     return MI->isAsCheapAsAMove();
 
   switch (MI->getOpcode()) {
   default:
     return false;
 
   // add/sub on register without shift
   case AArch64::ADDWri:

lib/Target/AArch64/AArch64LoadStoreOptimizer.cpp

for (MachineBasicBlock::iterator MBBI = MBB.begin(), E = MBB.end();
     }
     }
   }
 
   return Modified;
 }
 
 bool AArch64LoadStoreOpt::enableNarrowLdMerge(MachineFunction &Fn) {
-  bool ProfitableArch = Subtarget->isCortexA57();
+  bool ProfitableArch = Subtarget->isCortexA57() || Subtarget->isKryo();
   // FIXME: The benefit from converting narrow loads into a wider load could be
   // microarchitectural as it assumes that a single load with two bitfield
   // extracts is cheaper than two narrow loads. Currently, this conversion is
   // enabled only in cortex-a57 on which performance benefits were verified.
   return ProfitableArch && !Subtarget->requiresStrictAlign();
 }
 
 bool AArch64LoadStoreOpt::runOnMachineFunction(MachineFunction &Fn) {

lib/Target/AArch64/AArch64SchedKryo.td

+//==- AArch64SchedKryo.td - Qualcomm Kryo Scheduling Defs ---*- tablegen -*-==//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines the machine model for Qualcomm Kryo to support
+// instruction scheduling and other instruction cost heuristics.
+//
+//===----------------------------------------------------------------------===//
+
+//===----------------------------------------------------------------------===//
+// The issue witdh is set to five, matching the five issue queues for expanded
+// uops. Now, the latency spreadsheet has information based on fragmented uops,
+// but these do not actually take up an issue queue.
+
+def KryoModel : SchedMachineModel {
+  let IssueWidth        =   5; // 5-wide issue for expanded uops
+  let MicroOpBufferSize = 128; // Out-of-order with temporary unified issue buffer
+  let LoadLatency       =   4; // Optimistic load latency
+  let MispredictPenalty =  14; // Fetch + Decode/Rename/Dispatch + Branch
+
+  // Enable partial & runtime unrolling. The magic number is chosen based on
+  // experiments and benchmarking data.
+  let LoopMicroOpBufferSize = 16;
+}
+
+//===----------------------------------------------------------------------===//
+// Define each kind of processor resource and number available on Kryo.
+// Kryo has three different pipelines, each with a pair of pipes. However,
+// there are some limitations as to which of those individual pipes certain
+// instructions can go down. Therefore there is a AB group for each. Further,
+// some instructions can go down X or Y, so there's a XY group as well. There
+// are reservation stations X, Y, and LS; but we're not modelling them yet as
+// this will take some experimenting to get these reservations stations
+// modelled since they serve two pipe (A/B) each.
+//
+// Some instructions have a NoRSV uop which takes up an issue slot but does
+// not get booked. Therefore the NumMicroOps is higher than the size of the
+// ProcResource list. Additionally, some expanded uops are fragmented before
+// booking. The expanded uop uses an issues slot but the additional fragmented
+// uop does not. In this case the NumMicroOps is lower than the size of the
+// ProcResource list.
+
+let SchedModel = KryoModel in {
+  def KryoUnitXA : ProcResource<1>;                   // Type X(A) micro-ops
+  def KryoUnitXB : ProcResource<1>;                   // Type X(B) micro-ops
+  def KryoUnitYA : ProcResource<1>;                   // Type Y(A) micro-ops
+  def KryoUnitYB : ProcResource<1>;                   // Type Y(B) micro-ops
+  def KryoUnitX : ProcResGroup<[KryoUnitXA,          // Type X micro-ops
+                                KryoUnitXB]>;
+  def KryoUnitY : ProcResGroup<[KryoUnitYA,          // Type Y micro-ops
+                                KryoUnitYB]>;
+  def KryoUnitXY : ProcResGroup<[KryoUnitXA,         // Type XY micro-ops
+                                 KryoUnitXB,
+                                 KryoUnitYA,
+                                 KryoUnitYB]>;
+  def KryoUnitLSA : ProcResource<1>;                  // Type LS(A) micro-ops
+  def KryoUnitLSB : ProcResource<1>;                  // Type LS(B) micro-ops
+  def KryoUnitLS : ProcResGroup<[KryoUnitLSA,        // Type LS micro-ops
+                                 KryoUnitLSB]>;
+}
+
+let SchedModel = KryoModel in {
+
+//===----------------------------------------------------------------------===//
+// Map the target-defined scheduler read/write resources and latency for
+// Kryo.
+
+def : WriteRes<WriteImm,   [KryoUnitXY]> { let Latency = 1; }
+def : WriteRes<WriteI,     [KryoUnitXY]> { let Latency = 1; }
+def : WriteRes<WriteISReg, [KryoUnitXY, KryoUnitXY]>
+      { let Latency = 2; let NumMicroOps = 2; }
+def : WriteRes<WriteIEReg, [KryoUnitXY, KryoUnitXY]>
+      { let Latency = 2; let NumMicroOps = 2; }
+def : WriteRes<WriteExtr,  [KryoUnitXY, KryoUnitX]>
+      { let Latency = 2; let NumMicroOps = 2; }
+def : WriteRes<WriteIS,    [KryoUnitXY]> { let Latency = 2; }
+def : WriteRes<WriteID32,  [KryoUnitXA, KryoUnitY]>
+      { let Latency = 8; let NumMicroOps = 1; } // Fragent -1
+def : WriteRes<WriteID64,  [KryoUnitXA, KryoUnitY]>
+      { let Latency = 8; let NumMicroOps = 1; } // Fragent -1
+def : WriteRes<WriteIM32,  [KryoUnitX]> { let Latency = 5; }
+def : WriteRes<WriteIM64,  [KryoUnitX]> { let Latency = 5; }
+def : WriteRes<WriteBr,    [KryoUnitXY]> { let Latency = 1; }
+def : WriteRes<WriteBrReg, [KryoUnitXY]> { let Latency = 1; }
+def : WriteRes<WriteLD,    [KryoUnitLS]> { let Latency = 4; }
+def : WriteRes<WriteST,    [KryoUnitLS]> { let Latency = 4; }
+def : WriteRes<WriteSTP,   [KryoUnitLS]> { let Latency = 4; }
+def : WriteRes<WriteAdr,   [KryoUnitXY]> { let Latency = 6; }
+def : WriteRes<WriteLDIdx, [KryoUnitLS]> { let Latency = 4; }
+def : WriteRes<WriteSTIdx, [KryoUnitLS]> { let Latency = 4; }
+def : WriteRes<WriteF,     [KryoUnitXY, KryoUnitXY]>
+      { let Latency = 3; let NumMicroOps = 2; }
+def : WriteRes<WriteFCmp,  [KryoUnitXY]> { let Latency = 2; }
+def : WriteRes<WriteFCvt,  [KryoUnitX]> { let Latency = 4; }
+def : WriteRes<WriteFCopy, [KryoUnitXY]> { let Latency = 6; }
+def : WriteRes<WriteFImm,  [KryoUnitXY]> { let Latency = 6; }
+def : WriteRes<WriteFMul,  [KryoUnitX, KryoUnitX]>
+      { let Latency = 6; let NumMicroOps = 2; }
+def : WriteRes<WriteFDiv,  [KryoUnitXA, KryoUnitY]>
+      { let Latency = 12; let NumMicroOps = 2; } // Fragent -1 / NoRSV +1
+def : WriteRes<WriteV,     [KryoUnitXY]> { let Latency = 6; }
+def : WriteRes<WriteVLD,   [KryoUnitLS]> { let Latency = 4; }
+def : WriteRes<WriteVST,   [KryoUnitLS]> { let Latency = 4; }
+
+def : WriteRes<WriteSys,     []> { let Latency = 1; }
+def : WriteRes<WriteBarrier, []> { let Latency = 1; }
+def : WriteRes<WriteHint,    []> { let Latency = 1; }
+
+def : WriteRes<WriteLDHi,    []> { let Latency = 4; }
+
+// Nor forwarding logic is modelled yet.
+def : ReadAdvance<ReadI,       0>;
+def : ReadAdvance<ReadISReg,   0>;
+def : ReadAdvance<ReadIEReg,   0>;
+def : ReadAdvance<ReadIM,      0>;
+def : ReadAdvance<ReadIMA,     0>;
+def : ReadAdvance<ReadID,      0>;
+def : ReadAdvance<ReadExtrHi,  0>;
+def : ReadAdvance<ReadAdrBase, 0>;
+def : ReadAdvance<ReadVLD,     0>;
+
+
+//===----------------------------------------------------------------------===//
+// Specialize the coarse model by associating instruction groups with the
+// subtarget-defined types. As the modeled is refined, this will override most
+// of the above SchedWriteRes and SchedAlias mappings.
+
+// Miscellaneous
+// -----------------------------------------------------------------------------
+
+def : InstRW<[WriteI], (instrs COPY)>;
+
+
+// Detailed Refinedments
+// -----------------------------------------------------------------------------
+include "AArch64SchedKryoDetails.td"
+
+
+} // SchedModel = KryoModel

lib/Target/AArch64/AArch64SchedKryoDetails.td

+//=- AArch64SchedKryoDetails.td - QC Kryo Scheduling Defs ----*- tablegen -*-=//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines the uop and latency details for the machine model for the
+// Qualcomm Kryo subtarget.
+//
+// DO NOT MODIFY: This file it automatically generated.
+//
+//===----------------------------------------------------------------------===//
+
+def KryoWrite_3cyc_X_noRSV_138ln :
+  SchedWriteRes<[KryoUnitX]> {
+    let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_X_noRSV_138ln],
+    (instregex "(S|U)R?SRA(d|(v2i32|v4i16|v8i8)_shift)")>;
+
+def KryoWrite_3cyc_X_X_139ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+    let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_139ln],
+    (instregex "(S|U)R?SRA(v2i64|v4i32|v8i16|v16i8)_shift")>;
+
+def KryoWrite_4cyc_XY_XY_noRSV_172ln :
+    SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+    let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_XY_XY_noRSV_172ln],
+  (instregex "(S|U)ABA(v8i8|v4i16|v2i32)")>;
+def KryoWrite_4cyc_XY_XY_XY_XY_178ln :
+    SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitXY, KryoUnitXY]> {
+    let Latency = 4; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_4cyc_XY_XY_XY_XY_178ln],
+  (instregex "(S|U)ABA(v16i8|v8i16|v4i32)")>;
+def KryoWrite_3cyc_XY_XY_XY_XY_177ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitXY, KryoUnitXY]> {
+    let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_XY_XY_177ln],
+  (instregex "(S|U)ABALv.*")>;
+def KryoWrite_3cyc_XY_XY_166ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+    let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_166ln],
+  (instregex "(S|U)(ABD|QSUB|RHADD)(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_3cyc_XY_noRSV_159ln :
+  SchedWriteRes<[KryoUnitXY]> {
+    let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_159ln],
+  (instregex "(S|U)(ABD|RHADD)(v8i8|v4i16|v2i32)")>;
+def KryoWrite_3cyc_XY_XY_165ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+    let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_165ln],
+  (instregex "(S|U)ABDLv.*")>;
+def KryoWrite_3cyc_X_noRSV_154ln :
+  SchedWriteRes<[KryoUnitX]> {
+let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_X_noRSV_154ln],
+  (instregex "(S|U)ADALP(v8i8|v4i16|v2i32)_v.*")>;
+def KryoWrite_3cyc_X_X_155ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_155ln],
+  (instregex "(S|U)ADALP(v16i8|v8i16|v4i32)_v.*")>;
+def KryoWrite_2cyc_XY_XY_151ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_151ln],
+  (instregex "(S|U)(ADD|SUB)Lv.*")>;
+def KryoWrite_2cyc_XY_noRSV_148ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_148ln],
+  (instregex "((S|U)ADDLP|ABS)(v2i32|v4i16|v8i8)(_v.*)?")>;
+def KryoWrite_2cyc_XY_XY_150ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_150ln],
+  (instregex "((S|U)ADDLP|ABS)(v2i64|v4i32|v8i16|v16i8)(_v.*)?")>;
+def KryoWrite_3cyc_XY_XY_XY_noRSV_179ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_XY_noRSV_179ln],
+  (instrs SADDLVv4i32v, UADDLVv4i32v)>;
+def KryoWrite_5cyc_XY_XY_XY_noRSV_180ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitXY]> {
+  let Latency = 5; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_5cyc_XY_XY_XY_noRSV_180ln],
+  (instrs SADDLVv8i16v, UADDLVv8i16v)>;
+def KryoWrite_6cyc_XY_XY_X_noRSV_181ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_6cyc_XY_XY_X_noRSV_181ln],
+  (instrs SADDLVv16i8v, UADDLVv16i8v)>;
+def KryoWrite_3cyc_XY_noRSV_158ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_158ln],
+  (instrs SADDLVv4i16v, UADDLVv4i16v, ADDVv4i16v)>;
+def KryoWrite_4cyc_X_noRSV_169ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_169ln],
+  (instrs SADDLVv8i8v, UADDLVv8i8v, ADDVv8i8v)>;
+def KryoWrite_2cyc_XY_XY_XY_XY_176ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_XY_XY_176ln],
+  (instregex "(S|U)(ADDW|SUBW)v.*")>;
+def KryoWrite_4cyc_X_noRSV_40ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_40ln],
+  (instregex "(S|U)CVTFS(W|X)(D|S)ri")>;
+def KryoWrite_4cyc_X_noRSV_97ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_97ln],
+  (instregex "(S|U)CVTFU(W|X)(D|S)ri")>;
+def KryoWrite_4cyc_X_noRSV_110ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_110ln],
+  (instregex "(S|U)CVTF(v1i32|v2i32|v1i64|v2f32|d|s)(_shift)?")>;
+def KryoWrite_4cyc_X_X_114ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_114ln],
+  (instregex "(S|U)CVTF(v2i64|v4i32|v2f64|v4f32)(_shift)?")>;
+def KryoWrite_1cyc_XA_Y_98ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_98ln],
+  (instregex "(S|U)DIV(_Int)?(W|X)r")>;
+def KryoWrite_2cyc_XY_XY_152ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_152ln],
+  (instregex "(S|U)H(ADD|SUB)(v16i8|v8i16|v4i32)")>;
+def KryoWrite_2cyc_XY_noRSV_149ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_149ln],
+  (instregex "((S|U)H(ADD|SUB)|ADDP)(v8i8|v4i16|v2i32)")>;
+def KryoWrite_4cyc_X_70ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_X_70ln],
+  (instregex "(S|U)(MADDL|MSUBL)rrr")>;
+def KryoWrite_4cyc_X_X_191ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_191ln],
+  (instregex "(S|U|SQD)(MLAL|MLSL|MULL)v.*")>;
+def KryoWrite_1cyc_XY_195ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_195ln],
+  (instregex "(S|U)MOVv.*")>;
+def KryoWrite_5cyc_X_71ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_5cyc_X_71ln],
+  (instrs SMULHrr, UMULHrr)>;
+def KryoWrite_3cyc_XY_noRSV_186ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_186ln],
+  (instregex "^(S|U)QADD(v8i8|v4i16|v2i32)")>;
+def KryoWrite_3cyc_XY_XY_187ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_187ln],
+  (instregex "^(S|U)QADD(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_3cyc_XY_noRSV_69ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_69ln],
+  (instregex "(S|U|SU|US)QADD(v1i8|v1i16|v2i16|v1i32|v1i64)")>;
+def KryoWrite_3cyc_XY_noRSV_248ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_248ln],
+  (instregex "(S|U)QSHLU?(d|s|h|b|(v8i8|v4i16|v2i32)_shift)$")>;
+def KryoWrite_3cyc_XY_XY_250ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_250ln],
+  (instregex "(S|U)(QSHLU?|RSHR)(v16i8|v8i16|v4i32|v2i64)_shift$")>;
+def KryoWrite_3cyc_XY_noRSV_246ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_246ln],
+  (instregex "(S|U)(QSHL|RSHL|QRSHL)(v1i8|v1i16|v1i32|v1i64|v8i8|v4i16|v2i32)$")>;
+def KryoWrite_3cyc_XY_XY_251ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_251ln],
+  (instregex "(S|U)(QSHL|RSHL|QRSHL)(v16i8|v8i16|v4i32|v2i64)$")>;
+def KryoWrite_6cyc_XY_X_238ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_XY_X_238ln],
+  (instregex "((S|U)QR?SHRN|SQR?SHRUN)(v16i8|v8i16|v4i32)_shift$")>;
+def KryoWrite_3cyc_XY_noRSV_249ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_249ln],
+  (instregex "((S|U)QR?SHRN|SQR?SHRUN)(s|h|b)?")>;
+def KryoWrite_6cyc_XY_X_noRSV_252ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_6cyc_XY_X_noRSV_252ln],
+  (instregex "((S|U)QR?SHRN|SQR?SHRUN)(v8i8|v4i16|v2i32)_shift?")>;
+def KryoWrite_3cyc_XY_noRSV_161ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_161ln],
+  (instregex "(S|U)QSUB(v8i8|v4i16|v2i32|v1i64|v1i32|v1i16|v1i8)")>;
+def KryoWrite_3cyc_XY_noRSV_163ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_163ln],
+  (instregex "(S|U)QXTU?N(v16i8|v8i16|v4i32|v8i8|v4i16|v2i32)")>;
+def KryoWrite_3cyc_XY_noRSV_162ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_162ln],
+  (instregex "(S|U)QXTU?N(v1i8|v1i16|v1i32)")>;
+def KryoWrite_3cyc_XY_noRSV_247ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_247ln],
+  (instregex "(S|U)RSHR(d|(v8i8|v4i16|v2i32)_shift)$")>;
+def KryoWrite_2cyc_XY_noRSV_239ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_239ln],
+  (instregex "(S|U)SHL(d|v8i8|v4i16|v2i32|v1i64)$")>;
+def KryoWrite_2cyc_XY_XY_243ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_243ln],
+  (instregex "(S|U)SHL(v16i8|v8i16|v4i32|v2i64)$")>;
+def KryoWrite_2cyc_XY_XY_241ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_241ln],
+  (instregex "(S|U)?SHLL(v16i8|v8i16|v4i32|v8i8|v4i16|v2i32)(_shift)?$")>;
+def KryoWrite_2cyc_XY_noRSV_240ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_240ln],
+  (instregex "((S|U)SHR|SHL)(d|(v8i8|v4i16|v2i32)_shift)$")>;
+def KryoWrite_2cyc_XY_XY_242ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_242ln],
+  (instregex "((S|U)SHR|SHL)(v16i8|v8i16|v4i32|v2i64)_shift$")>;
+def KryoWrite_2cyc_XY_XY_183ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_183ln],
+  (instregex "(S|U)(MAX|MIN)P?(v16i8|v8i16|v4i32)")>;
+def KryoWrite_2cyc_XY_noRSV_182ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_182ln],
+  (instregex "(S|U)(MAX|MIN)P?(v8i8|v4i16|v2i32)")>;
+def KryoWrite_3cyc_XY_noRSV_184ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_184ln],
+  (instregex "(S|U)(MAX|MIN)V(v4i16v|v8i8v|v4i32)")>;
+def KryoWrite_4cyc_X_noRSV_185ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_185ln],
+  (instregex "(S|U)(MAX|MIN)V(v16i8v|v8i16v)")>;
+def KryoWrite_2cyc_XY_noRSV_67ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_67ln],
+  (instrs ABSv1i64)>;
+def KryoWrite_1cyc_XY_63ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_63ln, ReadI, ReadI],
+  (instregex "ADC.*")>;
+def KryoWrite_1cyc_XY_63_1ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_63_1ln],
+  (instregex "ADR.*")>;
+def KryoWrite_1cyc_XY_62ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_62ln, ReadI],
+  (instregex "ADDS?(W|X)ri")>;
+def KryoWrite_2cyc_XY_XY_64ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_64ln, ReadI, ReadI],
+  (instregex "ADDS?(W|X)r(r|s|x)(64)?")>;
+def KryoWrite_1cyc_XY_noRSV_65ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_65ln],
+  (instrs ADDv1i64)>;
+def KryoWrite_1cyc_XY_noRSV_144ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_144ln],
+  (instregex "(ADD|SUB)(v8i8|v4i16|v2i32|v1i64)")>;
+def KryoWrite_1cyc_XY_XY_146ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_146ln],
+  (instregex "(ADD|SUB)(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_4cyc_XY_X_noRSV_171ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_XY_X_noRSV_171ln],
+  (instregex "(ADD|SUB)HNv.*")>;
+def KryoWrite_1cyc_XY_noRSV_66ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_66ln],
+  (instrs ADDPv2i64p)>;
+def KryoWrite_2cyc_XY_XY_153ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_153ln],
+  (instregex "ADDP(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_3cyc_XY_XY_noRSV_170ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_noRSV_170ln],
+  (instrs ADDVv4i32v)>;
+def KryoWrite_4cyc_XY_XY_noRSV_173ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_XY_XY_noRSV_173ln],
+  (instrs ADDVv8i16v)>;
+def KryoWrite_5cyc_XY_X_noRSV_174ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_5cyc_XY_X_noRSV_174ln],
+  (instrs ADDVv16i8v)>;
+def KryoWrite_3cyc_XY_XY_X_X_27ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_X_X_27ln],
+  (instrs AESDrr, AESErr)>;
+def KryoWrite_2cyc_X_X_22ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_X_X_22ln],
+  (instrs AESIMCrr, AESMCrr)>;
+def KryoWrite_1cyc_XY_noRSV_76ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_76ln],
+  (instregex "((AND|ORN|EOR|EON)S?(Wr[rsi]|v8i8|v4i16|v2i32)|(ORR|BIC)S?(Wr[rs]|v8i8|v4i16|v2i32))")>;
+def KryoWrite_1cyc_XY_XY_79ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_79ln],
+  (instregex "((AND|ORN|EOR|EON)S?(Xr[rsi]|v16i8|v8i16|v4i32)|(ORR|BIC)S?(Xr[rs]|v16i8|v8i16|v4i32))")>;
+def KryoWrite_1cyc_X_72ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_X_72ln],
+  (instregex "(S|U)?BFM.*")>;
+def KryoWrite_1cyc_XY_noRSV_77ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_77ln],
+  (instregex "(BIC|ORR)S?Wri")>;
+def KryoWrite_1cyc_XY_XY_78ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_78ln],
+  (instregex "(BIC|ORR)S?Xri")>;
+def KryoWrite_1cyc_X_noRSV_74ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_noRSV_74ln],
+  (instrs BIFv8i8, BITv8i8, BSLv8i8)>;
+def KryoWrite_1cyc_X_X_75ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_75ln],
+  (instrs BIFv16i8, BITv16i8, BSLv16i8)>;
+def KryoWrite_0cyc_noRSV_11ln :
+  SchedWriteRes<[]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_noRSV_11ln],
+  (instrs BRK, DCPS1, DCPS2, DCPS3, HLT, HVC, ISB, HINT, SMC, SVC)>;
+def KryoWrite_0cyc_XY_16ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_XY_16ln, ReadI],
+  (instregex "(CCMN|CCMP)(W|X)i")>;
+def KryoWrite_0cyc_XY_16_1ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_XY_16_1ln, ReadI, ReadI],
+  (instregex "(CCMN|CCMP)(W|X)r")>;
+def KryoWrite_2cyc_XY_3ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_2cyc_XY_3ln, ReadI],
+  (instregex "(CLS|CLZ)(W|X)r")>;
+def KryoWrite_2cyc_XY_noRSV_7ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_7ln],
+  (instregex "(CLS|CLZ|CNT)(v4i32|v8i16|v16i8)")>;
+def KryoWrite_2cyc_XY_XY_8ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_8ln],
+  (instregex "(CLS|CLZ|CNT)(v2i32|v4i16|v8i8)")>;
+def KryoWrite_2cyc_XY_noRSV_80ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_80ln],
+  (instregex "CM(EQ|GE|HS|GT|HI|TST)(v8i8|v4i16|v2i32|v1i64)$")>;
+def KryoWrite_2cyc_XY_XY_83ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_83ln],
+  (instregex "CM(EQ|GE|HS|GT|HI|TST)(v16i8|v8i16|v4i32|v2i64)$")>;
+def KryoWrite_2cyc_XY_noRSV_81ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_81ln],
+  (instregex "CM(EQ|LE|GE|GT|LT)(v8i8|v4i16|v2i32|v1i64)rz$")>;
+def KryoWrite_2cyc_XY_XY_82ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_82ln],
+  (instregex "CM(EQ|LE|GE|GT|LT)(v16i8|v8i16|v4i32|v2i64)rz$")>;
+def KryoWrite_3cyc_XY_4ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_XY_4ln, ReadI, ReadISReg],
+  (instregex "CRC32.*")>;
+def KryoWrite_1cyc_XY_20ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_20ln, ReadI, ReadI],
+  (instregex "CSEL(W|X)r")>;
+def KryoWrite_1cyc_X_17ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_X_17ln, ReadI, ReadI],
+  (instregex "(CSINC|CSNEG)(W|X)r")>;
+def KryoWrite_1cyc_XY_18ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_18ln, ReadI, ReadI],
+  (instregex "(CSINV)(W|X)r")>;
+def KryoWrite_3cyc_LS_X_13ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_X_13ln],
+  (instrs DRPS)>;
+def KryoWrite_0cyc_LS_10ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_LS_10ln],
+  (instrs DSB, DMB, CLREX)>;
+def KryoWrite_1cyc_X_noRSV_196ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_noRSV_196ln],
+  (instregex "DUP(v8i8|v4i16|v2i32)(gpr|lane)")>;
+def KryoWrite_1cyc_X_X_197ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_197ln],
+  (instregex "DUP(v16i8|v8i16|v4i32|v2i64)(gpr|lane)")>;
+def KryoWrite_3cyc_LS_LS_X_15ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_X_15ln],
+  (instrs ERET)>;
+def KryoWrite_1cyc_X_noRSV_207ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_noRSV_207ln],
+  (instrs EXTv8i8)>;
+def KryoWrite_1cyc_X_X_212ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_212ln],
+  (instrs EXTv16i8)>;
+def KryoWrite_2cyc_XY_X_136ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_X_136ln],
+  (instrs EXTRWrri, EXTRXrri)>;
+def KryoWrite_2cyc_XY_noRSV_35ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_35ln],
+  (instregex "F(MAX|MIN)(NM)?P?(D|S)rr")>;
+def KryoWrite_2cyc_XY_XY_106ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_106ln],
+  (instregex "(F(MAX|MIN)(NM)?P?|FAC(GE|GT)|FCM(EQ|GE|GT))(v2i64p|v2f64|v4f32)")>;
+def KryoWrite_2cyc_XY_noRSV_104ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_104ln],
+  (instregex "(F(MAX|MIN)(NM)?P?|FAC(GE|GT)|FCM(EQ|GE|GT))(v2f32|v2i32p)")>;
+def KryoWrite_3cyc_XY_noRSV_107ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_107ln],
+  (instregex "F(MAX|MIN)(NM)?Vv4i32v")>;
+def KryoWrite_3cyc_XY_noRSV_101ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_101ln],
+  (instregex "FABD(32|64|v2f32)")>;
+def KryoWrite_3cyc_XY_XY_103ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_103ln],
+  (instregex "(FABD|FADD|FSUB|FADDP)(v4f32|v2f64)")>;
+def KryoWrite_1cyc_XY_noRSV_48ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_48ln],
+  (instregex "F(ABS|NEG)(D|S)r")>;
+def KryoWrite_1cyc_XY_noRSV_124ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_124ln],
+  (instregex "F(ABS|NEG)v2f32")>;
+def KryoWrite_1cyc_XY_XY_125ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_125ln],
+  (instregex "F(ABS|NEG)(v2f64|v4f32)")>;
+def KryoWrite_2cyc_XY_noRSV_33ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_33ln],
+  (instregex "(FAC(GE|GT)|FCM(EQ|GE|GT))(32|64)")>;
+def KryoWrite_3cyc_XY_noRSV_30ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_30ln],
+  (instregex "(FADD|FSUB)(D|S)rr")>;
+def KryoWrite_3cyc_XY_noRSV_100ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_100ln],
+  (instregex "(FADD|FSUB|FADDP)v2f32")>;
+def KryoWrite_3cyc_XY_noRSV_29ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_29ln],
+  (instregex "FADDP(v2i32p|v2i64p)")>;
+def KryoWrite_0cyc_XY_31ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_XY_31ln],
+  (instregex "FCCMPE?(D|S)rr")>;
+def KryoWrite_2cyc_XY_noRSV_34ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_34ln],
+  (instregex "FCM(EQ|LE|GE|GT|LT)(v1i32|v1i64)rz")>;
+def KryoWrite_2cyc_XY_XY_36ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_36ln],
+  (instregex "FCM(EQ|LE|GE|GT|LT)(v2i64|v4i32)rz")>;
+def KryoWrite_2cyc_XY_noRSV_105ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_105ln],
+  (instregex "FCM(EQ|LE|GE|GT|LT)v2i32rz")>;
+def KryoWrite_0cyc_XY_32ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_XY_32ln],
+  (instregex "FCMPE?(D|S)r(r|i)")>;
+def KryoWrite_1cyc_XY_noRSV_49ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_49ln],
+  (instrs FCSELDrrr, FCSELSrrr)>;
+def KryoWrite_4cyc_X_noRSV_41ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_41ln],
+  (instrs FCVTDHr, FCVTDSr, FCVTHDr, FCVTHSr, FCVTSDr, FCVTSHr)>;
+def KryoWrite_4cyc_X_38ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_X_38ln],
+  (instregex "FCVT(((A|N|M|P)(S|U)(S|U)|Z(S|U)_Int(S|U))(W|X)(D|S)ri?|Z(S|U)(d|s))$")>;
+def KryoWrite_4cyc_X_noRSV_113ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_113ln],
+  (instregex "FCVT((A|N|M|P)(S|U)|Z(S|U)_Int)(v1i32|v1i64|v2f32)$")>;
+def KryoWrite_4cyc_X_X_117ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_117ln],
+  (instregex "FCVT((A|N|M|P)(S|U)|Z(S|U)_Int)(v4f32|v2f64)$")>;
+def KryoWrite_5cyc_X_X_XY_noRSV_119ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitXY]> {
+  let Latency = 5; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_XY_noRSV_119ln],
+  (instregex "FCVTX?N(v2f32|v4f32|v2i32|v4i16|v4i32|v8i16)$")>;
+def KryoWrite_4cyc_X_X_116ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_116ln],
+  (instregex "FCVTL(v2i32|v4i16|v4i32|v8i16)$")>;
+def KryoWrite_4cyc_X_noRSV_112ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_112ln],
+  (instrs FCVTXNv1i64)>;
+def KryoWrite_4cyc_X_37ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_X_37ln],
+  (instregex "FCVTZ(S|U)(S|U)(W|X)(D|S)ri?$")>;
+def KryoWrite_4cyc_X_noRSV_111ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_111ln],
+  (instregex "FCVTZ(S|U)(v2f32|v1i32|v1i64|v2i32(_shift)?)$")>;
+def KryoWrite_4cyc_X_X_115ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_115ln],
+  (instregex "FCVTZ(S|U)(v2f64|v4f32|(v2i64|v4i32)(_shift)?)$")>;
+def KryoWrite_1cyc_XA_Y_noRSV_43ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_noRSV_43ln],
+  (instrs FDIVDrr, FDIVSrr)>;
+def KryoWrite_1cyc_XA_Y_noRSV_121ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_noRSV_121ln],
+  (instrs FDIVv2f32)>;
+def KryoWrite_1cyc_XA_Y_XA_Y_123ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY, KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_XA_Y_123ln],
+  (instrs FDIVv2f64, FDIVv4f32)>;
+def KryoWrite_5cyc_X_noRSV_55ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_55ln],
+  (instregex "FN?M(ADD|SUB)Srrr")>;
+def KryoWrite_6cyc_X_noRSV_57ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_noRSV_57ln],
+  (instregex "FN?M(ADD|SUB)Drrr")>;
+def KryoWrite_5cyc_X_noRSV_51ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_51ln],
+  (instrs FMLAv2f32, FMLSv2f32, FMLAv1i32_indexed, FMLSv1i32_indexed)>;
+def KryoWrite_5cyc_X_X_56ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_56ln],
+  (instrs FMLAv4f32, FMLSv4f32)>;
+def KryoWrite_6cyc_X_X_61ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_X_61ln],
+  (instrs FMLAv2f64, FMLSv2f64)>;
+def KryoWrite_5cyc_X_noRSV_128ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_128ln],
+  (instrs FMLAv2i32_indexed, FMLSv2i32_indexed)>;
+def KryoWrite_5cyc_X_X_131ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_131ln],
+  (instrs FMLAv4i32_indexed, FMLSv4i32_indexed)>;
+def KryoWrite_6cyc_X_X_134ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_X_134ln],
+  (instrs FMLAv2i64_indexed, FMLSv2i64_indexed)>;
+def KryoWrite_6cyc_X_noRSV_60ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_noRSV_60ln],
+  (instrs FMLAv1i64_indexed, FMLSv1i64_indexed, FMULv1i64_indexed, FMULXv1i64_indexed)>;
+def KryoWrite_1cyc_XY_45ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_45ln],
+  (instregex "FMOV(XDHigh|DXHigh|DX)r")>;
+def KryoWrite_1cyc_XY_noRSV_47ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_47ln],
+  (instregex "FMOV(Di|Dr|Si|Sr|SWr|WSr|XDr|v.*_ns)")>;
+def KryoWrite_5cyc_X_noRSV_53ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_53ln],
+  (instrs FMULv1i32_indexed, FMULXv1i32_indexed)>;
+def KryoWrite_5cyc_X_noRSV_127ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_127ln],
+  (instrs FMULv2f32, FMULXv2f32, FMULv2i32_indexed, FMULXv2i32_indexed)>;
+def KryoWrite_5cyc_X_X_130ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_130ln],
+  (instrs FMULv4f32, FMULXv4f32, FMULv4i32_indexed, FMULXv4i32_indexed)>;
+def KryoWrite_6cyc_X_X_133ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_X_133ln],
+  (instrs FMULv2f64, FMULXv2f64, FMULv2i64_indexed, FMULXv2i64_indexed)>;
+def KryoWrite_5cyc_X_noRSV_54ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_54ln],
+  (instrs FMULSrr, FNMULSrr, FMULX32)>;
+def KryoWrite_6cyc_X_noRSV_59ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_noRSV_59ln],
+  (instrs FMULDrr, FNMULDrr, FMULX64)>;
+def KryoWrite_3cyc_XY_noRSV_28ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_28ln],
+  (instrs FRECPEv1i32, FRECPEv1i64, FRSQRTEv1i32, FRSQRTEv1i64 )>;
+def KryoWrite_3cyc_XY_noRSV_99ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_99ln],
+  (instrs FRECPEv2f32, FRSQRTEv2f32)>;
+def KryoWrite_3cyc_XY_XY_102ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_102ln],
+  (instrs FRECPEv2f64, FRECPEv4f32, FRSQRTEv2f64, FRSQRTEv4f32)>;
+def KryoWrite_5cyc_X_noRSV_52ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_52ln],
+  (instrs FRECPS32, FRSQRTS32)>;
+def KryoWrite_6cyc_X_noRSV_58ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_noRSV_58ln],
+  (instrs FRECPS64, FRSQRTS64)>;
+def KryoWrite_5cyc_X_noRSV_126ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_noRSV_126ln],
+  (instrs FRECPSv2f32, FRSQRTSv2f32)>;
+def KryoWrite_5cyc_X_X_129ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_129ln],
+  (instrs FRECPSv4f32, FRSQRTSv4f32)>;
+def KryoWrite_6cyc_X_X_132ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_6cyc_X_X_132ln],
+  (instrs FRECPSv2f64, FRSQRTSv2f64)>;
+def KryoWrite_3cyc_XY_noRSV_50ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_50ln],
+  (instrs FRECPXv1i32, FRECPXv1i64)>;
+def KryoWrite_2cyc_XY_noRSV_39ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_39ln],
+  (instregex "FRINT(A|I|M|N|P|X|Z)(S|D)r")>;
+def KryoWrite_2cyc_XY_noRSV_108ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_108ln],
+  (instregex "FRINT(A|I|M|N|P|X|Z)v2f32")>;
+def KryoWrite_2cyc_XY_XY_109ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_109ln],
+  (instregex "FRINT(A|I|M|N|P|X|Z)(v2f64|v4f32)")>;
+def KryoWrite_1cyc_XA_Y_noRSV_42ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_noRSV_42ln],
+  (instregex "FSQRT(S|D)r")>;
+def KryoWrite_1cyc_XA_Y_noRSV_120ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_noRSV_120ln],
+  (instregex "FSQRTv2f32")>;
+def KryoWrite_1cyc_XA_Y_XA_Y_122ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitY, KryoUnitXA, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_1cyc_XA_Y_XA_Y_122ln],
+  (instregex "FSQRT(v2f64|v4f32)")>;
+def KryoWrite_1cyc_X_201ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_X_201ln],
+  (instregex "INSv.*")>;
+def KryoWrite_3cyc_LS_255ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_255ln],
+  (instregex "LD1(One(v16b|v8h|v4s|v2d)|i64)$")>;
+def KryoWrite_4cyc_LS_X_270ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_LS_X_270ln],
+  (instregex "LD1(i8|i16|i32)$")>;
+def KryoWrite_3cyc_LS_noRSV_285ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_285ln],
+  (instregex "LD1One(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_289ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_289ln, WriteAdr],
+  (instregex "LD1(One(v16b|v8h|v4s|v2d)|i64)_POST$")>;
+def KryoWrite_4cyc_LS_XY_X_298ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_X_298ln, WriteAdr],
+  (instregex "LD1(i8|i16|i32)_POST$")>;
+def KryoWrite_3cyc_LS_LS_LS_308ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_LS_308ln],
+  (instregex "LD1Three(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_XY_noRSV_317ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_noRSV_317ln, WriteAdr],
+  (instregex "LD1One(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_LS_LS_328ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_LS_LS_328ln, WriteAdr],
+  (instregex "LD1Four(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_XY_LS_LS_332ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_LS_332ln, WriteAdr],
+  (instregex "LD1Three(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_348ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_348ln],
+  (instregex "LD1Three(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_LS_LS_351ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_LS_LS_351ln],
+  (instregex "LD1Four(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_noRSV_358ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_noRSV_358ln],
+  (instregex "LD1Four(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_360ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_360ln, WriteAdr],
+  (instregex "LD1Three(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_noRSV_368ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 7;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_noRSV_368ln, WriteAdr],
+  (instregex "LD1Four(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_281ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_281ln],
+  (instregex "LD(1|2)Two(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_noRSV_noRSV_311ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_noRSV_311ln],
+  (instregex "LD(1|2)Two(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_313ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_313ln, WriteAdr],
+  (instregex "LD(1|2)Two(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_XY_noRSV_noRSV_334ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_noRSV_noRSV_334ln, WriteAdr],
+  (instregex "LD(1|2)Two(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_256ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_256ln],
+  (instregex "LD1R(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_noRSV_286ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_286ln],
+  (instregex "LD1R(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_290ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_290ln, WriteAdr],
+  (instregex "LD1R(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_XY_noRSV_318ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_noRSV_318ln, WriteAdr],
+  (instregex "LD1R(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_257ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_257ln],
+  (instregex "LD2i64$")>;
+def KryoWrite_3cyc_LS_XY_291ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_291ln, WriteAdr],
+  (instregex "LD2i64_POST$")>;
+def KryoWrite_4cyc_LS_X_X_296ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_LS_X_X_296ln],
+  (instregex "LD2(i8|i16|i32)$")>;
+def KryoWrite_4cyc_LS_XY_X_X_321ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_X_X_321ln, WriteAdr],
+  (instregex "LD2(i8|i16|i32)_POST$")>;
+def KryoWrite_3cyc_LS_LS_282ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_282ln],
+  (instregex "LD2R(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_noRSV_noRSV_312ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_noRSV_312ln],
+  (instregex "LD2R(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_314ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_314ln, WriteAdr],
+  (instregex "LD2R(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_XY_noRSV_noRSV_335ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_noRSV_noRSV_335ln, WriteAdr],
+  (instregex "LD2R(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_283ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_283ln],
+  (instregex "LD3i64$")>;
+def KryoWrite_3cyc_LS_LS_LS_309ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_LS_309ln],
+  (instregex "LD3Threev2d$")>;
+def KryoWrite_3cyc_LS_XY_LS_315ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_315ln, WriteAdr],
+  (instregex "LD3i64_POST$")>;
+def KryoWrite_4cyc_LS_X_X_X_320ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_4cyc_LS_X_X_X_320ln],
+  (instregex "LD3(i8|i16|i32)$")>;
+def KryoWrite_3cyc_LS_XY_LS_LS_331ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_LS_331ln, WriteAdr],
+  (instregex "LD3Threev2d_POST$")>;
+def KryoWrite_4cyc_LS_XY_X_X_X_338ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_X_X_X_338ln, WriteAdr],
+  (instregex "LD3(i8|i16|i32)_POST$")>;
+def KryoWrite_4cyc_LS_LS_X_X_X_noRSV_noRSV_noRSV_373ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 8;
+}
+def : InstRW<[KryoWrite_4cyc_LS_LS_X_X_X_noRSV_noRSV_noRSV_373ln],
+  (instregex "LD3Three(v8b|v4h|v2s)$")>;
+def KryoWrite_4cyc_LS_XY_LS_X_X_X_noRSV_noRSV_noRSV_380ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitX, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 9;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_LS_X_X_X_noRSV_noRSV_noRSV_380ln, WriteAdr],
+  (instregex "LD3Three(v8b|v4h|v2s)_POST$")>;
+def KryoWrite_4cyc_LS_LS_X_X_X_LS_LS_X_X_X_381ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 10;
+}
+def : InstRW<[KryoWrite_4cyc_LS_LS_X_X_X_LS_LS_X_X_X_381ln],
+  (instregex "LD3Three(v16b|v8h|v4s)$")>;
+def KryoWrite_4cyc_LS_LS_X_X_X_LS_XY_LS_X_X_X_383ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitX, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 11;
+}
+def : InstRW<[KryoWrite_4cyc_LS_LS_X_X_X_LS_XY_LS_X_X_X_383ln, WriteAdr],
+  (instregex "LD3Three(v16b|v8h|v4s)_POST$")>;
+def KryoWrite_3cyc_LS_LS_LS_310ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_LS_310ln],
+  (instregex "LD3R(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_LS_333ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_LS_333ln, WriteAdr],
+  (instregex "LD3R(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_349ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_349ln],
+  (instregex "LD3R(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_361ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_361ln, WriteAdr],
+  (instregex "LD3R(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_284ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_284ln],
+  (instregex "LD4i64$")>;
+def KryoWrite_3cyc_LS_XY_LS_316ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_316ln, WriteAdr],
+  (instregex "LD4i64_POST$")>;
+def KryoWrite_3cyc_LS_LS_LS_LS_329ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_LS_LS_329ln],
+  (instregex "LD4Four(v2d)$")>;
+def KryoWrite_4cyc_LS_X_X_X_X_337ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_4cyc_LS_X_X_X_X_337ln],
+  (instregex "LD4(i8|i16|i32)$")>;
+def KryoWrite_3cyc_LS_XY_LS_LS_LS_350ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_LS_LS_350ln, WriteAdr],
+  (instregex "LD4Four(v2d)_POST$")>;
+def KryoWrite_4cyc_LS_XY_X_X_X_X_355ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_X_X_X_X_355ln, WriteAdr],
+  (instregex "LD4(i8|i16|i32)_POST$")>;
+def KryoWrite_4cyc_LS_LS_X_X_X_X_noRSV_noRSV_noRSV_noRSV_382ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 10;
+}
+def : InstRW<[KryoWrite_4cyc_LS_LS_X_X_X_X_noRSV_noRSV_noRSV_noRSV_382ln],
+  (instregex "LD4Four(v8b|v4h|v2s)$")>;
+def KryoWrite_4cyc_LS_XY_LS_X_X_X_X_noRSV_noRSV_noRSV_noRSV_384ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 11;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_LS_X_X_X_X_noRSV_noRSV_noRSV_noRSV_384ln, WriteAdr],
+  (instregex "LD4Four(v8b|v4h|v2s)_POST$")>;
+def KryoWrite_4cyc_LS_LS_X_X_X_X_LS_LS_X_X_X_X_386ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 12;
+}
+def : InstRW<[KryoWrite_4cyc_LS_LS_X_X_X_X_LS_LS_X_X_X_X_386ln],
+  (instregex "LD4Four(v16b|v8h|v4s)$")>;
+def KryoWrite_4cyc_LS_LS_X_X_X_X_LS_XY_LS_X_X_X_X_389ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 13;
+}
+def : InstRW<[KryoWrite_4cyc_LS_LS_X_X_X_X_LS_XY_LS_X_X_X_X_389ln, WriteAdr],
+  (instregex "LD4Four(v16b|v8h|v4s)_POST$")>;
+def KryoWrite_3cyc_LS_LS_LS_LS_330ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_LS_LS_330ln],
+  (instregex "LD4R(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_LS_LS_352ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS, KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_LS_LS_352ln, WriteAdr],
+  (instregex "LD4R(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_noRSV_359ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_noRSV_noRSV_noRSV_noRSV_359ln],
+  (instregex "LD4R(v8b|v4h|v2s|v1d)$")>;
+def KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_noRSV_369ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 7;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_noRSV_noRSV_noRSV_noRSV_369ln, WriteAdr],
+  (instregex "LD4R(v8b|v4h|v2s|v1d)_POST$")>;
+def KryoWrite_3cyc_LS_LS_400ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_400ln],
+  (instregex "(LDAX?R(B|H|W|X)|LDAXP(W|X))")>;
+def KryoWrite_3cyc_LS_LS_401ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_401ln, WriteLDHi],
+  (instrs LDNPQi)>;
+def KryoWrite_3cyc_LS_noRSV_noRSV_408ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_noRSV_408ln, WriteLDHi],
+  (instrs LDNPDi, LDNPSi)>;
+def KryoWrite_3cyc_LS_394ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_394ln, WriteLDHi],
+  (instrs LDNPWi, LDNPXi)>;
+def KryoWrite_3cyc_LS_LS_402ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_402ln, WriteLDHi],
+  (instrs LDPQi)>;
+def KryoWrite_3cyc_LS_noRSV_noRSV_409ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_noRSV_409ln, WriteLDHi],
+  (instrs LDPDi, LDPSi)>;
+def KryoWrite_3cyc_LS_XY_LS_410ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY, KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_LS_410ln, WriteLDHi, WriteAdr],
+  (instregex "LDPQ(post|pre)")>;
+def KryoWrite_3cyc_LS_XY_noRSV_noRSV_411ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_noRSV_noRSV_411ln, WriteLDHi, WriteAdr],
+  (instregex "LDP(D|S)(post|pre)")>;
+def KryoWrite_3cyc_LS_393ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_393ln, WriteLDHi],
+  (instrs LDPWi, LDPXi)>;
+def KryoWrite_3cyc_LS_XY_403ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_403ln, WriteLDHi, WriteAdr],
+  (instregex "LDP(W|X)(post|pre)")>;
+def KryoWrite_4cyc_LS_395ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_LS_395ln, WriteLDHi],
+  (instrs LDPSWi)>;
+def KryoWrite_4cyc_LS_XY_405ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_405ln, WriteLDHi, WriteAdr],
+  (instrs LDPSWpost, LDPSWpre)>;
+def KryoWrite_3cyc_LS_264ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_264ln],
+  (instrs LDRQui, LDRQl)>;
+def KryoWrite_4cyc_X_LS_271ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_LS_271ln],
+  (instrs LDRQroW, LDRQroX)>;
+def KryoWrite_3cyc_LS_noRSV_287ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_287ln],
+  (instregex "LDR((D|S)l|(D|S|H|B)ui)")>;
+def KryoWrite_3cyc_LS_XY_293ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_293ln, WriteAdr],
+  (instrs LDRQpost, LDRQpre)>;
+def KryoWrite_4cyc_X_LS_noRSV_297ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_X_LS_noRSV_297ln],
+  (instregex "LDR(D|S|H|B)ro(W|X)")>;
+def KryoWrite_3cyc_LS_XY_noRSV_319ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_noRSV_319ln, WriteAdr],
+  (instregex "LDR(D|S|H|B)(post|pre)")>;
+def KryoWrite_3cyc_LS_261ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_261ln],
+  (instregex "LDR(BB|HH|W|X)ui")>;
+def KryoWrite_3cyc_LS_XY_292ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_XY_292ln, WriteAdr],
+  (instregex "LDR(BB|HH|W|X)(post|pre)")>;
+def KryoWrite_4cyc_X_LS_272ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_LS_272ln],
+  (instregex "(LDR(BB|HH|W|X)ro(W|X)|PRFMro(W|X))")>;
+def KryoWrite_3cyc_LS_262ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_262ln],
+  (instrs LDRWl, LDRXl)>;
+def KryoWrite_4cyc_LS_268ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_LS_268ln],
+  (instregex "LDRS(BW|BX|HW|HX|W)ui")>;
+def KryoWrite_5cyc_X_LS_273ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS]> {
+  let Latency = 5; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_5cyc_X_LS_273ln],
+  (instregex "LDRS(BW|BX|HW|HX|W)ro(W|X)")>;
+def KryoWrite_4cyc_LS_XY_294ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitXY]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_LS_XY_294ln, WriteAdr],
+  (instregex "LDRS(BW|BX|HW|HX|W)(post|pre)")>;
+def KryoWrite_4cyc_LS_269ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_LS_269ln],
+  (instrs LDRSWl)>;
+def KryoWrite_3cyc_LS_260ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_260ln],
+  (instregex "LDTR(B|H|W|X)i")>;
+def KryoWrite_4cyc_LS_267ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_LS_267ln],
+  (instregex "LDTRS(BW|BX|HW|HX|W)i")>;
+def KryoWrite_3cyc_LS_263ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_263ln],
+  (instrs LDURQi)>;
+def KryoWrite_3cyc_LS_noRSV_288ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_noRSV_288ln],
+  (instregex "LDUR(D|S|H|B)i")>;
+def KryoWrite_3cyc_LS_259ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_259ln],
+  (instregex "LDUR(BB|HH|W|X)i")>;
+def KryoWrite_4cyc_LS_266ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_LS_266ln],
+  (instregex "LDURS(B|H)?(W|X)i")>;
+def KryoWrite_3cyc_LS_258ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_258ln],
+  (instregex "LDXP(W|X)")>;
+def KryoWrite_3cyc_LS_258_1ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 3; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_3cyc_LS_258_1ln],
+  (instregex "LDXR(B|H|W|X)")>;
+def KryoWrite_2cyc_XY_XY_137ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_137ln],
+  (instrs LSLVWr, LSLVXr)>;
+def KryoWrite_1cyc_XY_135ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_135ln],
+  (instregex "(LS|AS|RO)RV(W|X)r")>;
+def KryoWrite_4cyc_X_84ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_4cyc_X_84ln],
+  (instrs MADDWrrr, MSUBWrrr)>;
+def KryoWrite_5cyc_X_85ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_5cyc_X_85ln],
+  (instrs MADDXrrr, MSUBXrrr)>;
+def KryoWrite_4cyc_X_noRSV_188ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_188ln],
+  (instregex "(MLA|MLS|MUL)(v8i8|v4i16|v2i32)(_indexed)?")>;
+def KryoWrite_4cyc_X_X_192ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_192ln],
+  (instregex "(MLA|MLS|MUL|SQR?DMULH)(v16i8|v8i16|v4i32)(_indexed)?")>;
+def KryoWrite_1cyc_XY_noRSV_198ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_198ln],
+  (instregex "(MOVI|MVNI)(D|v8b_ns|v2i32|v4i16|v2s_msl)")>;
+def KryoWrite_1cyc_XY_XY_199ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_199ln],
+  (instregex "(MOVI|MVNI)(v2d_ns|v16b_ns|v4i32|v8i16|v4s_msl)")>;
+def KryoWrite_1cyc_X_89ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_X_89ln],
+  (instrs MOVKWi, MOVKXi)>;
+def KryoWrite_1cyc_XY_91ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_91ln],
+  (instrs MOVNWi, MOVNXi)>;
+def KryoWrite_1cyc_XY_90ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_90ln],
+  (instrs MOVZWi, MOVZXi)>;
+def KryoWrite_2cyc_XY_93ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_2cyc_XY_93ln],
+  (instrs MRS)>;
+def KryoWrite_0cyc_X_87ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_X_87ln],
+  (instrs MSRpstateImm4)>;
+def : InstRW<[KryoWrite_0cyc_X_87ln],
+  (instrs MSRpstateImm1)>;
+def KryoWrite_0cyc_XY_88ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_XY_88ln],
+  (instrs MSR)>;
+def KryoWrite_1cyc_XY_noRSV_143ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_143ln],
+  (instregex "NEG(v8i8|v4i16|v2i32|v1i64)")>;
+def KryoWrite_1cyc_XY_XY_145ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_145ln],
+  (instregex "NEG(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_1cyc_XY_noRSV_193ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_193ln],
+  (instrs NOTv8i8)>;
+def KryoWrite_1cyc_XY_XY_194ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_194ln],
+  (instrs NOTv16i8)>;
+def KryoWrite_2cyc_XY_noRSV_234ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_234ln],
+  (instrs PMULv8i8)>;
+def KryoWrite_2cyc_XY_XY_236ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_236ln],
+  (instrs PMULv16i8)>;
+def KryoWrite_2cyc_XY_XY_235ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_235ln],
+  (instrs PMULLv8i8, PMULLv16i8)>;
+def KryoWrite_3cyc_XY_XY_237ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_237ln],
+  (instrs PMULLv1i64, PMULLv2i64)>;
+def KryoWrite_0cyc_LS_254ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_LS_254ln],
+  (instrs PRFMl, PRFMui)>;
+def KryoWrite_0cyc_LS_253ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_LS_253ln],
+  (instrs PRFUMi)>;
+def KryoWrite_6cyc_XY_X_noRSV_175ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitX]> {
+  let Latency = 6; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_6cyc_XY_X_noRSV_175ln],
+  (instregex "R(ADD|SUB)HNv.*")>;
+def KryoWrite_2cyc_XY_204ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_2cyc_XY_204ln],
+  (instrs RBITWr, RBITXr)>;
+def KryoWrite_2cyc_XY_noRSV_218ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_noRSV_218ln],
+  (instrs RBITv8i8)>;
+def KryoWrite_2cyc_XY_XY_219ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_219ln],
+  (instrs RBITv16i8)>;
+def KryoWrite_1cyc_X_202ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_X_202ln],
+  (instregex "REV(16|32)?(W|X)r")>;
+def KryoWrite_1cyc_XY_noRSV_214ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_214ln],
+  (instregex "REV(16|32|64)(v8i8|v4i16|v2i32)")>;
+def KryoWrite_1cyc_XY_XY_216ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_216ln],
+  (instregex "REV(16|32|64)(v16i8|v8i16|v4i32)")>;
+def KryoWrite_3cyc_X_noRSV_244ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_X_noRSV_244ln],
+  (instregex "S(L|R)I(d|(v8i8|v4i16|v2i32)_shift)")>;
+def KryoWrite_3cyc_X_X_245ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_245ln],
+  (instregex "S(L|R)I(v16i8|v8i16|v4i32|v2i64)_shift")>;
+def KryoWrite_1cyc_XY_2ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_2ln, ReadI, ReadI],
+  (instregex "SBCS?(W|X)r")>;
+def KryoWrite_2cyc_XA_XA_XA_24ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitXA, KryoUnitXA]> {
+  let Latency = 2; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_2cyc_XA_XA_XA_24ln],
+  (instrs SHA1Crrr, SHA1Mrrr, SHA1Prrr)>;
+def KryoWrite_1cyc_XY_noRSV_21ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_21ln],
+  (instrs SHA1Hrr)>;
+def KryoWrite_2cyc_X_X_23ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_X_X_23ln],
+  (instrs SHA1SU0rrr, SHA1SU1rr, SHA256SU0rr)>;
+def KryoWrite_4cyc_XA_XA_XA_25ln :
+  SchedWriteRes<[KryoUnitXA, KryoUnitXA, KryoUnitXA]> {
+  let Latency = 4; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_4cyc_XA_XA_XA_25ln],
+  (instrs SHA256Hrrr, SHA256H2rrr)>;
+def KryoWrite_3cyc_XY_XY_X_X_26ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY, KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_X_X_26ln],
+  (instrs SHA256SU1rrr)>;
+def KryoWrite_4cyc_X_noRSV_189ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_189ln],
+  (instregex "SQR?DMULH(v8i8|v4i16|v1i32|v2i32|v1i16)(_indexed)?")>;
+def KryoWrite_3cyc_XY_noRSV_68ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_68ln],
+  (instregex "SQ(ABS|NEG)(v1i8|v1i16|v1i32|v1i64)")>;
+def KryoWrite_3cyc_XY_noRSV_157ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_157ln],
+  (instregex "SQ(ABS|NEG)(v8i8|v4i16|v2i32)")>;
+def KryoWrite_3cyc_XY_XY_164ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_164ln],
+  (instregex "SQ(ABS|NEG)(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_4cyc_X_noRSV_190ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_4cyc_X_noRSV_190ln],
+  (instregex "SQD(MLAL|MLSL|MULL)(i16|i32)")>;
+def KryoWrite_0cyc_LS_Y_274ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_274ln],
+  (instregex "ST1(One(v8b|v4h|v2s|v1d|v16b|v8h|v4s|v2d)|(i8|i16|i32|i64)|Two(v8b|v4h|v2s|v1d))$")>;
+def KryoWrite_1cyc_LS_Y_X_301ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_X_301ln],
+  (instregex "ST1(One(v8b|v4h|v2s|v1d|v16b|v8h|v4s|v2d)|(i8|i16|i32|i64)|Two(v8b|v4h|v2s|v1d))_POST$")>;
+def KryoWrite_1cyc_LS_Y_XY_305ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_305ln],
+  (instregex "ST1(One(v16b|v8h|v4s|v2d)|Two(v8b|v4h|v2s|v1d))_POST$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_323ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 4;
+}
+def : InstRW<[WriteAdr, KryoWrite_0cyc_LS_Y_LS_Y_323ln],
+  (instregex "ST1(Two(v16b|v8h|v4s|v2d)|(Three|Four)(v8b|v4h|v2s|v1d))_POST$")>;
+def KryoWrite_1cyc_LS_Y_XY_LS_Y_345ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_1cyc_LS_Y_XY_LS_Y_345ln],
+  (instregex "ST1(Two(v16b|v8h|v4s|v2d)|(Three|Four)(v8b|v4h|v2s|v1d))$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_356ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_356ln],
+  (instregex "ST1Three(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_1cyc_LS_Y_XY_LS_Y_LS_Y_366ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS, KryoUnitY,
+                   KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 7;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_LS_Y_LS_Y_366ln],
+  (instregex "ST1Three(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_LS_Y_371ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 8;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_LS_Y_371ln],
+  (instregex "ST1Four(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_XY_LS_Y_LS_Y_377ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY, KryoUnitXY,
+                   KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 9;
+}
+def : InstRW<[WriteAdr, KryoWrite_0cyc_LS_Y_LS_Y_XY_LS_Y_LS_Y_377ln],
+  (instregex "ST1Four(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_0cyc_LS_Y_275ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_275ln],
+  (instregex "ST2(Two(v8b|v4h|v2s|v1d|v16b|v8h|v4s|v2d)|(i8|i16|i32|i64))$")>;
+def KryoWrite_1cyc_LS_Y_XY_306ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_306ln],
+  (instregex "ST2(Two(v8b|v4h|v2s|v1d)|(i8|i16|i32|i64))_POST$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_322ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_322ln],
+  (instregex "ST2Two(v16b|v8h|v4s|v2d)$")>;
+def KryoWrite_1cyc_LS_Y_XY_LS_Y_344ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 5;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_LS_Y_344ln],
+  (instregex "ST2Two(v16b|v8h|v4s|v2d)_POST$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_324ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_324ln],
+  (instregex "ST3(Threev1d|(i8|i16|i32|i64))$")>;
+def KryoWrite_1cyc_LS_Y_XY_LS_Y_346ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 5;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_LS_Y_346ln],
+  (instregex "ST3(Threev1d|(i8|i16|i32|i64))_POST$")>;
+def KryoWrite_1cyc_X_X_LS_Y_LS_Y_353ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_LS_Y_LS_Y_353ln],
+  (instregex "ST3Three(v8b|v4h|v2s)$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_357ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_357ln],
+  (instregex "ST3Threev2d$")>;
+def KryoWrite_1cyc_X_X_LS_Y_XY_LS_Y_363ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitXY,
+                   KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 7;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_X_X_LS_Y_XY_LS_Y_363ln],
+  (instregex "ST3Three(v8b|v4h|v2s)_POST$")>;
+def KryoWrite_1cyc_LS_Y_XY_LS_Y_LS_Y_367ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS, KryoUnitY,
+                   KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 7;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_LS_Y_LS_Y_367ln],
+  (instregex "ST3Threev2d_POST$")>;
+def KryoWrite_1cyc_X_X_LS_Y_LS_Y_X_X_LS_Y_LS_Y_385ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY, KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY,
+                   KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 12;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_LS_Y_LS_Y_X_X_LS_Y_LS_Y_385ln],
+  (instregex "ST3Three(v16b|v8h|v4s)$")>;
+def KryoWrite_1cyc_X_X_LS_Y_LS_Y_X_X_LS_Y_XY_LS_Y_388ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY, KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY,
+                   KryoUnitXY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 13;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_X_X_LS_Y_LS_Y_X_X_LS_Y_XY_LS_Y_388ln],
+  (instregex "ST3Three(v16b|v8h|v4s)_POST$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_325ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_325ln],
+  (instregex "ST4(Fourv1d|(i8|i16|i32|i64))$")>;
+def KryoWrite_1cyc_LS_Y_XY_LS_Y_347ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 5;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_XY_LS_Y_347ln],
+  (instregex "ST4(Fourv1d|(i8|i16|i32|i64))_POST$")>;
+def KryoWrite_1cyc_X_X_LS_Y_X_X_LS_Y_370ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitX,
+                   KryoUnitX, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 8;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_LS_Y_X_X_LS_Y_370ln],
+  (instregex "ST4Four(v8b|v4h|v2s)$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_LS_Y_372ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY, KryoUnitLS,
+                   KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 8;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_LS_Y_LS_Y_372ln],
+  (instregex "ST4Fourv2d$")>;
+def KryoWrite_1cyc_X_X_LS_Y_XY_X_X_LS_Y_375ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitXY,
+                   KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 9;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_X_X_LS_Y_XY_X_X_LS_Y_375ln],
+  (instregex "ST4Four(v8b|v4h|v2s)_POST$")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_XY_LS_Y_LS_Y_379ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY, KryoUnitXY,
+                   KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 9;
+}
+def : InstRW<[WriteAdr, KryoWrite_0cyc_LS_Y_LS_Y_XY_LS_Y_LS_Y_379ln],
+  (instregex "ST4Fourv2d_POST$")>;
+def KryoWrite_1cyc_X_X_LS_Y_X_X_LS_Y_X_X_LS_Y_X_X_LS_Y_390ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitX,
+                   KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitX, KryoUnitX,
+                   KryoUnitLS, KryoUnitY, KryoUnitX, KryoUnitX, KryoUnitLS,
+                   KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 16;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_LS_Y_X_X_LS_Y_X_X_LS_Y_X_X_LS_Y_390ln],
+  (instregex "ST4Four(v16b|v8h|v4s)$")>;
+def KryoWrite_1cyc_X_X_LS_Y_X_X_LS_Y_X_X_LS_Y_XY_X_X_LS_Y_392ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitX,
+                   KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitX, KryoUnitX,
+                   KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitX, KryoUnitX,
+                   KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 17;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_X_X_LS_Y_X_X_LS_Y_X_X_LS_Y_XY_X_X_LS_Y_392ln],
+  (instregex "ST4Four(v16b|v8h|v4s)_POST$")>;
+def KryoWrite_0cyc_LS_LS_Y_299ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_0cyc_LS_LS_Y_299ln],
+  (instregex "STLR(B|H|W|X)")>;
+def KryoWrite_3cyc_LS_LS_Y_307ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitLS, KryoUnitY]> {
+  let Latency = 3; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_3cyc_LS_LS_Y_307ln],
+  (instregex "STLX(P(W|X)|R(B|H|W|X))")>;
+def KryoWrite_0cyc_LS_Y_276ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_276ln],
+  (instrs STNPDi, STNPSi)>;
+def KryoWrite_0cyc_LS_Y_LS_Y_326ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_326ln],
+  (instrs STNPQi)>;
+def KryoWrite_0cyc_LS_Y_280ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_280ln],
+  (instrs STNPWi, STNPXi)>;
+def KryoWrite_0cyc_LS_Y_277ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_277ln],
+  (instregex "STP(D|S)i")>;
+def KryoWrite_1cyc_LS_Y_X_303ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_X_303ln],
+  (instregex "STP(D|S)(post|pre)")>;
+def KryoWrite_0cyc_LS_Y_LS_Y_327ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_LS_Y_327ln],
+  (instrs STPQi)>;
+def KryoWrite_1cyc_LS_Y_X_LS_Y_343ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitX, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 5;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_X_LS_Y_343ln],
+  (instrs STPQpost, STPQpre)>;
+def KryoWrite_0cyc_LS_Y_279ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_279ln],
+  (instregex "STP(W|X)i")>;
+def KryoWrite_1cyc_LS_X_Y_300ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_X_Y_300ln],
+  (instregex "STP(W|X)(post|pre)")>;
+def KryoWrite_0cyc_LS_Y_278ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_278ln],
+  (instregex "STR(Q|D|S|H|B)ui")>;
+def KryoWrite_1cyc_X_LS_Y_295ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_1cyc_X_LS_Y_295ln],
+  (instregex "STR(D|S|H|B)ro(W|X)")>;
+def KryoWrite_1cyc_LS_Y_X_304ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_Y_X_304ln],
+  (instregex "STR(Q|D|S|H|B)(post|pre)")>;
+def KryoWrite_2cyc_X_LS_Y_XY_LS_Y_354ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS, KryoUnitY, KryoUnitXY, KryoUnitLS,
+                   KryoUnitY]> {
+  let Latency = 2; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_2cyc_X_LS_Y_XY_LS_Y_354ln],
+  (instregex "STRQro(W|X)")>;
+def KryoWrite_0cyc_LS_Y_399ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_399ln],
+  (instregex "STR(BB|HH|W|X)ui")>;
+def KryoWrite_1cyc_X_LS_Y_406ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitLS, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_1cyc_X_LS_Y_406ln],
+  (instregex "STR(BB|HH|W|X)ro(W|X)")>;
+def KryoWrite_1cyc_LS_X_Y_407ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitX, KryoUnitY]> {
+  let Latency = 1; let NumMicroOps = 3;
+}
+def : InstRW<[WriteAdr, KryoWrite_1cyc_LS_X_Y_407ln],
+  (instregex "STR(BB|HH|W|X)(post|pre)")>;
+def KryoWrite_0cyc_LS_Y_398ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_398ln],
+  (instregex "STTR(B|H|W|X)i")>;
+def KryoWrite_0cyc_LS_Y_396ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_396ln],
+  (instregex "STUR(Q|D|S|H|B)i")>;
+def KryoWrite_0cyc_LS_Y_397ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 0; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_0cyc_LS_Y_397ln],
+  (instregex "STUR(BB|HH|W|X)i")>;
+def KryoWrite_3cyc_LS_Y_404ln :
+  SchedWriteRes<[KryoUnitLS, KryoUnitY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_LS_Y_404ln],
+  (instregex "STX(P(W|X)|R(B|H|W|X))")>;
+def KryoWrite_3cyc_XY_noRSV_160ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_160ln],
+  (instregex "^(SU|US)QADD(v8i8|v4i16|v2i32)")>;
+def KryoWrite_3cyc_XY_XY_167ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_167ln],
+  (instregex "^(SU|US)QADD(v16i8|v8i16|v4i32|v2i64)")>;
+def KryoWrite_1cyc_XY_1ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_1cyc_XY_1ln, ReadI],
+  (instregex "SUBS?(W|X)ri")>;
+def KryoWrite_2cyc_XY_XY_5ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_5ln, ReadI, ReadIEReg],
+  (instregex "SUBS?(W|X)rx")>;
+def KryoWrite_2cyc_XY_XY_5_1ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 2; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_2cyc_XY_XY_5_1ln, ReadI, ReadISReg],
+  (instregex "SUBS?(W|X)rs")>;
+def KryoWrite_1cyc_XY_noRSV_6ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_6ln, ReadI, ReadI],
+  (instregex "SUBS?(W|X)rr")>;
+def KryoWrite_0cyc_LS_9ln :
+  SchedWriteRes<[KryoUnitLS]> {
+  let Latency = 0; let NumMicroOps = 1;
+}
+def : InstRW<[KryoWrite_0cyc_LS_9ln],
+  (instregex "SYSL?xt")>;
+def KryoWrite_1cyc_X_noRSV_205ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_noRSV_205ln],
+  (instrs TBLv8i8One)>;
+def KryoWrite_1cyc_X_X_208ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_208ln],
+  (instrs TBLv16i8One)>;
+def KryoWrite_2cyc_X_X_X_noRSV_222ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_2cyc_X_X_X_noRSV_222ln],
+  (instrs TBLv8i8Two)>;
+def KryoWrite_2cyc_X_X_X_X_X_X_224ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_2cyc_X_X_X_X_X_X_224ln],
+  (instrs TBLv16i8Two)>;
+def KryoWrite_3cyc_X_X_X_X_X_noRSV_225ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 6;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_X_X_X_noRSV_225ln],
+  (instrs TBLv8i8Three)>;
+def KryoWrite_3cyc_X_X_X_X_X_X_X_noRSV_228ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 8;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_X_X_X_X_X_noRSV_228ln],
+  (instrs TBLv8i8Four)>;
+def KryoWrite_4cyc_X_X_X_X_X_X_X_X_XY_X_X_230ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitXY, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 11;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_X_X_X_X_X_X_XY_X_X_230ln],
+  (instrs TBLv16i8Three)>;
+def KryoWrite_4cyc_X_X_X_X_X_X_X_X_X_X_XY_X_X_X_X_232ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitXY, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 15;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_X_X_X_X_X_X_X_X_XY_X_X_X_X_232ln],
+  (instrs TBLv16i8Four)>;
+def KryoWrite_2cyc_X_X_noRSV_220ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 3;
+}
+def : InstRW<[KryoWrite_2cyc_X_X_noRSV_220ln],
+  (instrs TBXv8i8One)>;
+def KryoWrite_2cyc_X_X_X_X_221ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 2; let NumMicroOps = 4;
+}
+def : InstRW<[KryoWrite_2cyc_X_X_X_X_221ln],
+  (instrs TBXv16i8One)>;
+def KryoWrite_3cyc_X_X_X_X_noRSV_223ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 5;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_X_X_noRSV_223ln],
+  (instrs TBXv8i8Two)>;
+def KryoWrite_4cyc_X_X_X_X_X_X_noRSV_226ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 7;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_X_X_X_X_noRSV_226ln],
+  (instrs TBXv8i8Three)>;
+def KryoWrite_3cyc_X_X_X_X_X_X_X_X_227ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 3; let NumMicroOps = 8;
+}
+def : InstRW<[KryoWrite_3cyc_X_X_X_X_X_X_X_X_227ln],
+  (instrs TBXv16i8Two)>;
+def KryoWrite_4cyc_X_X_X_X_X_X_X_X_noRSV_229ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 4; let NumMicroOps = 9;
+}
+def : InstRW<[KryoWrite_4cyc_X_X_X_X_X_X_X_X_noRSV_229ln],
+  (instrs TBXv8i8Four)>;
+def KryoWrite_5cyc_X_X_X_X_X_X_X_X_X_XY_X_X_X_231ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitXY,
+                   KryoUnitX, KryoUnitX, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 13;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_X_X_X_X_X_X_X_XY_X_X_X_231ln],
+  (instrs TBXv16i8Three)>;
+def KryoWrite_5cyc_X_X_X_X_X_X_X_X_X_X_X_XY_X_X_X_X_X_233ln :
+    SchedWriteRes<[KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitXY, KryoUnitX, KryoUnitX, KryoUnitX,
+                   KryoUnitX, KryoUnitX]> {
+  let Latency = 5; let NumMicroOps = 17;
+}
+def : InstRW<[KryoWrite_5cyc_X_X_X_X_X_X_X_X_X_X_X_XY_X_X_X_X_X_233ln],
+  (instrs TBXv16i8Four)>;
+def KryoWrite_1cyc_XY_XY_217ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_XY_217ln],
+  (instregex "((TRN1|TRN2|ZIP1|UZP1|UZP2)v2i64|ZIP2(v2i64|v4i32|v8i16|v16i8))")>;
+def KryoWrite_1cyc_X_X_211ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_211ln],
+  (instregex "(TRN1|TRN2)(v4i32|v8i16|v16i8)")>;
+def KryoWrite_1cyc_X_XY_213ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_XY_213ln],
+  (instregex "(TRN1|TRN2)(v2i32|v4i16|v8i8)")>;
+def KryoWrite_3cyc_XY_noRSV_156ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_noRSV_156ln],
+  (instrs URECPEv2i32, URSQRTEv2i32)>;
+def KryoWrite_3cyc_XY_XY_168ln :
+  SchedWriteRes<[KryoUnitXY, KryoUnitXY]> {
+  let Latency = 3; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_3cyc_XY_XY_168ln],
+  (instrs URECPEv4i32, URSQRTEv4i32)>;
+def KryoWrite_1cyc_X_X_210ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_210ln],
+  (instregex "(UZP1|UZP2)(v4i32|v8i16|v16i8)")>;
+def KryoWrite_1cyc_X_noRSV_206ln :
+  SchedWriteRes<[KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_noRSV_206ln],
+  (instregex "(UZP1|UZP2|ZIP1|ZIP2)(v2i32|v4i16|v8i8)")>;
+def KryoWrite_1cyc_XY_noRSV_215ln :
+  SchedWriteRes<[KryoUnitXY]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_XY_noRSV_215ln],
+  (instregex "XTNv.*")>;
+def KryoWrite_1cyc_X_X_209ln :
+  SchedWriteRes<[KryoUnitX, KryoUnitX]> {
+  let Latency = 1; let NumMicroOps = 2;
+}
+def : InstRW<[KryoWrite_1cyc_X_X_209ln],
+  (instregex "ZIP1(v4i32|v8i16|v16i8)")>;

lib/Target/AArch64/AArch64Subtarget.h

 class AArch64Subtarget : public AArch64GenSubtargetInfo {
 protected:
   enum ARMProcFamilyEnum {
     Others,
     CortexA35,
     CortexA53,
     CortexA57,
     Cyclone,
-    ExynosM1
+    ExynosM1,
+    Kryo
   };
 
   /// ARMProcFamily - ARM processor family: Cortex-A53, Cortex-A57, and others.
   ARMProcFamilyEnum ARMProcFamily;
 
   bool HasV8_1aOps;
   bool HasV8_2aOps;
 
public:
   bool isTargetELF() const { return TargetTriple.isOSBinFormatELF(); }
   bool isTargetMachO() const { return TargetTriple.isOSBinFormatMachO(); }
 
   bool isGeneric() const { return CPUString == "generic"; }
   bool isCyclone() const { return CPUString == "cyclone"; }
   bool isCortexA57() const { return CPUString == "cortex-a57"; }
   bool isCortexA53() const { return CPUString == "cortex-a53"; }
   bool isExynosM1() const { return CPUString == "exynos-m1"; }
+  bool isKryo() const { return CPUString == "kryo"; }
 
   bool useAA() const override { return isCortexA53(); }
 
   /// getMaxInlineSizeThreshold - Returns the maximum memset / memcpy size
   /// that still makes it profitable to inline the call.
   unsigned getMaxInlineSizeThreshold() const { return 64; }
 
   /// ParseSubtargetFeatures - Parses features string setting specified

lib/Target/AArch64/AArch64TargetTransformInfo.cpp

for (auto *I : Tys) {
     if (I->getScalarSizeInBits() * I->getVectorNumElements() == 128)
       Cost += getMemoryOpCost(Instruction::Store, I, 128, 0) +
         getMemoryOpCost(Instruction::Load, I, 128, 0);
   }
   return Cost;
 }
 
 unsigned AArch64TTIImpl::getMaxInterleaveFactor(unsigned VF) {
-  if (ST->isCortexA57())
+  if (ST->isCortexA57() || ST->isKryo())
     return 4;
   return 2;
 }
 
 void AArch64TTIImpl::getUnrollingPreferences(Loop *L,
                                              TTI::UnrollingPreferences &UP) {
   // Enable partial unrolling and runtime unrolling.
   BaseT::getUnrollingPreferences(L, UP);

test/CodeGen/AArch64/cpus.ll

 ; This tests that llc accepts all valid AArch64 CPUs
 
 
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=generic 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=cortex-a35 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=cortex-a53 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=cortex-a57 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=cortex-a72 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=exynos-m1 2>&1 | FileCheck %s
+; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=kryo 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=invalidcpu 2>&1 | FileCheck %s --check-prefix=INVALID
 
 ; CHECK-NOT: {{.*}}  is not a recognized processor for this target
 ; INVALID: {{.*}}  is not a recognized processor for this target
 
 define i32 @f(i64 %z) {
   ret i32 0
 }

test/CodeGen/AArch64/remat.ll

 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=cortex-a35 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=cortex-a57 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=cortex-a53 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=cortex-a72 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=exynos-m1 -o - %s | FileCheck %s
+; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=kryo -o - %s | FileCheck %s
 
 %X = type { i64, i64, i64 }
 declare void @f(%X*)
 define void @t() {
 entry:
   %tmp = alloca %X
   call void @f(%X* %tmp)
 ; CHECK: add x0, sp, #8
 ; CHECK-NEXT-NOT: mov
   call void @f(%X* %tmp)               
 ; CHECK: add x0, sp, #8
 ; CHECK-NEXT-NOT: mov
   ret void 
 }